- Refreshed patches: 55 (systemtap), 102 (lib64), 111 (no static lib),
114 (statvfs flags), 132 (unittest rpmbuild hooks), 134 (fix COUNT_ALLOCS in
test_sys), 143 (tsc on ppc64), 146 (hashlib fips), 153 (test gdb noise),
157 (UID+GID overflows), 173 (ENOPROTOOPT in bind_port), 186 (dont raise
from py_compile)
- Removed patches: 129 (test_subprocess nonreadable dir - no longer fails in
Koji), 142 (the mock issue that caused this is fixed)
- Added patch 187 (remove thread atfork) - will be in next version
- Refreshed script for checking pyc and pyo timestamps with new ignored files.
- The fips patch is disabled for now until upstream makes a final decision
what to do with sha3 implementation for 3.4.0.
- Refreshed patches: 55 (systemtap), 111 (no static lib), 146 (hashlib fips),
153 (fix test_gdb noise), 157 (uid, gid overflow - fixed upstream, just
keeping few more downstream tests)
- Removed patches: 3 (audiotest.au made it to upstream tarball)
- Removed workaround for http://bugs.python.org/issue14774, discussed in
http://bugs.python.org/issue15298 and fixed in revision 24d52d3060e8.
* Mon Mar 25 2013 David Malcolm <dmalcolm@redhat.com> - 3.3.0-9
- renumber patches to keep them in sync with python.spec
Specifically, renumber:
00174 -> 00176
00175 -> 00177
* Mon Aug 13 2012 David Malcolm <dmalcolm@redhat.com> - 3.3.0-0.3.b2
- 3.3b1 -> 3.3b2; drop upstreamed patch 152; refresh patches 3, 102, 111,
134, 153, 160; regenenerate autotools patch; rework systemtap patch to work
correctly when LANG=C (patch 55); importlib.test was moved to
test.test_importlib upstream
* Tue May 15 2012 David Malcolm <dmalcolm@redhat.com> - 3.2.3-7
- update uid/gid handling to avoid int overflows seen with uid/gid
values >= 2^31 on 32-bit architectures (patch 157; rhbz#697470)
* Fri Apr 20 2012 David Malcolm <dmalcolm@redhat.com> - 3.2.3-3
- add explicit version requirements on expat to avoid linkage problems with
XML_SetHashSalt
* Mon Oct 10 2011 David Malcolm <dmalcolm@redhat.com> - 3.2.2-8
- cherrypick fix for distutils not using __pycache__ when byte-compiling
files (rhbz#722578)
* Wed Sep 14 2011 David Malcolm <dmalcolm@redhat.com> - 3.2.2-5
- support OpenSSL FIPS mode in _hashlib and hashlib; don't build the _md5 and
_sha* modules, relying on _hashlib in hashlib (rhbz#563986; patch 146)
* Mon Sep 12 2011 David Malcolm <dmalcolm@redhat.com> - 3.2.2-3
- renumber and rename patches for consistency with python.spec (8 to 55, 106
to 104, 6 to 111, 104 to 113, 105 to 114, 125, 131, 130 to 143)
* Sat Sep 10 2011 David Malcolm <dmalcolm@redhat.com> - 3.2.2-2
- rewrite of "check", introducing downstream-only hooks for skipping specific
cases in an rpmbuild (patch 132), and fixing/skipping failing tests in a more
fine-grained manner than before; (patches 106, 133-142 sparsely, moving
patches for consistency with python.spec: 128 to 134, 126 to 135, 127 to 141)
- set EXTRA_CFLAGS to our CFLAGS, rather than overriding OPT, fixing a linker
error with dynamic annotations (when configured using --with-valgrind)
- fix the ppc build of the debug configuration (patch 130; rhbz#661510)
- 3.2b2
- rework patch 3 (removal of mimeaudio tests), patch 6 (no static libs),
patch 8 (systemtap), patch 102 (lib64)
- remove patch 4 (rendered redundant by upstream r85537), patch 103 (PEP 3149),
patch 110 (upstreamed expat fix), patch 111 (parallel build fix for grammar
fixed upstream)
- regenerate patch 300 (autotool intermediates)
- workaround COUNT_ALLOCS weakref issues in test suite (patch 126, patch 127,
patch 128)
- stop using runtest.sh in %%check (dropped by upstream), replacing with
regrtest; fixup list of failing tests
- introduce "pyshortver", "SOABI_optimized" and "SOABI_debug" macros
- rework manifests of shared libraries to use "SOABI_" macros, reflecting
PEP 3149
- drop itertools, operator and _collections modules from the manifests as py3k
commit r84058 moved these inside libpython; json/tests moved to test/json_tests
- move turtle code into the tkinter subpackage
* Sat Aug 21 2010 David Malcolm <dmalcolm@redhat.com> - 3.2-0.0.a1
- 3.2a1; add alphatag
- rework %%files in the light of PEP 3147 (__pycache__)
- drop our configuration patch to Setup.dist (patch 0): setup.py should do a
better job of things, and the %%files explicitly lists our modules (r82746
appears to break the old way of doing things). This leads to various modules
changing from "foomodule.so" to "foo.so". It also leads to the optimized build
dropping the _sha1, _sha256 and _sha512 modules, but these are provided by
_hashlib; _weakref becomes a builtin module; xxsubtype goes away (it's only for
testing/devel purposes)
- fixup patches 3, 4, 6, 8, 102, 103, 105, 111 for the rebase
- remove upstream patches: 7 (system expat), 106, 107, 108 (audioop reformat
plus CVE-2010-1634 and CVE-2010-2089), 109 (CVE-2008-5983)
- add machinery for rebuilding "configure" and friends, using the correct
version of autoconf (patch 300)
- patch the debug build's usage of COUNT_ALLOCS to be less verbose (patch 125)
- "modulator" was removed upstream
- drop "-b" from patch applications affecting .py files to avoid littering the
installation tree
- reformat whitespace in audioop.c (patch 106)
- CVE-2010-1634: fix various integer overflow checks in the audioop module
(patch 107)
- CVE-2010-2089: further checks within the audioop module (patch 108)
- CVE-2008-5983: the new PySys_SetArgvEx entry point from r81399 (patch
109)
- drop upstreamed patch 2 (.pyc permissions handling)
- drop upstream patch 5 (fix for the test_tk and test_ttk_* selftests)
- drop upstreamed patch 200 (path-fixing script)
- add systemtap static markers (wcohen, mjw, dmalcolm; patch 8), a
systemtap tapset defining "python.function.entry" and
"python.function.return" to make the markers easy to use (dmalcolm;
source 5), and an example of using the tapset to the docs (dmalcolm;
source 6) (rhbz:545179)
%%{_prefix}/lib/debug/%%{_libdir}/INSTSONAME.debug-gdb.py to avoid
noise from ldconfig (bug 562980), and which should also ensure it
becomes part of the debuginfo subpackage, rather than the libs
subpackage
- introduce %%{py_SOVERSION} and %%{py_INSTSONAME} to reflect the upstream
configure script, and to avoid fragile scripts that try to figure this
out dynamically (e.g. for the -gdb.py change)
build of pyexpat and elementtree modules
- add patch adapted from upstream (patch 7) to add support for building
against system expat; add --with-system-expat to "configure" invocation
- remove embedded copies of expat and zlib from source tree during "prep"
- explicitly list all lib-dynload files, rather than dynamically gathering
the payload into a temporary text file, so that we can be sure what we
are shipping
- introduce a macros.pybytecompile source file, to help with packaging
python3 modules (Source3; written by Toshio)
- rename "2to3-3" to "python3-2to3" to better reflect python 3 module
packaging plans
curses configuration in Modules/Setup.dist, so that the curses modules
are built using setup.py with the downstream default (linking against
libncursesw.so, rather than libncurses.so), rather than within the
Makefile; add a test to %%install to verify the dso files that the
curses module is linked against the correct DSO (bug 539917; changes
_cursesmodule.so -> _curses.so)
